linux-kernel - Re: [PATCH 05/10] arm64: mm: Re-implement the __tlbi

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <404625f2-808a-4578-98bd-4d12d7a72c7c@arm.com>
Date: Mon, 14 Jul 2025 10:02:34 +0100
From: Ryan Roberts <ryan.roberts@....com>
To: Will Deacon <will@...nel.org>, linux-arm-kernel@...ts.infradead.org
Cc: linux-kernel@...r.kernel.org, Ard Biesheuvel <ardb@...nel.org>,
 Catalin Marinas <catalin.marinas@....com>,
 Mark Rutland <mark.rutland@....com>,
 Linus Torvalds <torvalds@...ux-foundation.org>,
 Oliver Upton <oliver.upton@...ux.dev>, Marc Zyngier <maz@...nel.org>
Subject: Re: [PATCH 05/10] arm64: mm: Re-implement the __tlbi_level macro in C

On 11/07/2025 17:17, Will Deacon wrote:
> __tlbi_level() is just a simple macro around __tlbi_level_op(), so merge
> the two into a single C function. Drop the redundant comparison of
> 'u32 level' against 0 and tidy up the code a little while we're at it.
> 
> Signed-off-by: Will Deacon <will@...nel.org>
> ---
>  arch/arm64/include/asm/tlbflush.h | 25 ++++++++++---------------
>  1 file changed, 10 insertions(+), 15 deletions(-)
> 
> diff --git a/arch/arm64/include/asm/tlbflush.h b/arch/arm64/include/asm/tlbflush.h
> index 728b00f3e1f4..ddd77e92b268 100644
> --- a/arch/arm64/include/asm/tlbflush.h
> +++ b/arch/arm64/include/asm/tlbflush.h
> @@ -128,8 +128,17 @@ enum tlbi_op {
>  	___GEN_TLBI_OP_CASE(op);					\
>  		break
>  
> -static __always_inline void __tlbi_level_op(const enum tlbi_op op, u64 arg)
> +static __always_inline void __tlbi_level(const enum tlbi_op op, u64 addr, u32 level)

level is passed into all the higher level functions as int. That's why the
"level >= 0" comparison was previously there. Of course no users should be
calling this with a negative value so I'll guess that I was trying to guard
against seeing a valid level -1 in future with 6 level/4K D128 pgtables.

Given everything else uses signed int for level, perhaps we should be consistent
here too?

>  {
> +	u64 arg = addr;
> +
> +	if (alternative_has_cap_unlikely(ARM64_HAS_ARMv8_4_TTL) && level <= 3) {
> +		u64 ttl = level | (get_trans_granule() << 2);
> +
> +		arg &= ~TLBI_TTL_MASK;
> +		arg |= FIELD_PREP(TLBI_TTL_MASK, ttl);
> +	}
> +
>  	switch (op) {
>  	__GEN_TLBI_OP_ASID_CASE(vae1is);
>  	__GEN_TLBI_OP_CASE(vae2is);
> @@ -145,20 +154,6 @@ static __always_inline void __tlbi_level_op(const enum tlbi_op op, u64 arg)
>  #undef __GEN_TLBI_OP_ASID_CASE
>  #undef ___GEN_TLBI_OP_CASE
>  
> -#define __tlbi_level(op, addr, level) do {				\
> -	u64 arg = addr;							\
> -									\
> -	if (alternative_has_cap_unlikely(ARM64_HAS_ARMv8_4_TTL) &&	\
> -	    level >= 0 && level <= 3) {					\
> -		u64 ttl = level & 3;					\
> -		ttl |= get_trans_granule() << 2;			\
> -		arg &= ~TLBI_TTL_MASK;					\
> -		arg |= FIELD_PREP(TLBI_TTL_MASK, ttl);			\
> -	}								\
> -									\
> -	__tlbi_level_op(op, arg);					\
> -} while(0)
> -
>  /*
>   * This macro creates a properly formatted VA operand for the TLB RANGE. The
>   * value bit assignments are: